IPTV QoE: telemetry and user experience on Android TV set-top boxes

An IPTV operator with Android TV set-top boxes (Dboxes) distributed across multiple regions faced the classic video operations problem: everything looked "fine" from the data center, yet users kept reporting slowness, freezing and channels that would not load. The operation had backbone network and server metrics, but no visibility into the actual state of the device inside the home or the perceived quality when changing channels.

THE CHALLENGE

Dboxes are closed devices that, once installed in the customer's home, are difficult to diagnose remotely. When a user reports an issue, the operations center has no objective evidence: it does not know whether the cause is the home WiFi, the video app, an overloaded content server, an overheated set-top box or a combination of factors.

Diagnosis was done by elimination, based on phone calls or chats, without reproducible measurements. This extended mean time to resolution, generated unnecessary technical visits and left the operation permanently reactive.

THE SOLUTION

We designed and implemented a telemetry and quality-of-experience (QoE) platform made up of field rigs and a central manager. The goal was to measure what the user actually experiences, not what the infrastructure assumes.

Field rigs

Each rig is installed on the same local network as the Dboxes. When powered on, it automatically discovers devices via adb-over-WiFi, provisions them with the required support APKs and adjusts power, animation and logging settings so telemetry remains stable. Rig workers run telemetry and QoE tasks published from the central manager, using task queues and per-device locks to avoid collisions.

Continuous device telemetry

The platform periodically captures the real state of each Dbox:

  • Network: latency, jitter and packet loss toward multiple endpoints measured from the set-top box itself.
  • System: CPU load, RAM usage, disk usage and top processes.
  • WiFi: received signal strength (RSSI), signal-to-noise ratio (SNR), negotiated link speed, effective throughput and internal quality score reported by the device.
  • Thermal: processor temperature and throttling level applied by the system.
  • System events: structured logs of network, playback, app lifecycle and system stability.

Real QoE tests

In addition to passive telemetry, the platform runs active tests that mimic the user's real interaction. It opens the video app, measures the time until the stream starts playing, zaps through configured channels and records stalls, rebuffers, dropped frames, bitrates, achieved resolution and program metadata. Everything runs on the customer's actual hardware, not in a lab.

Central manager

The central manager aggregates information, presents live dashboards with progressive updates, geolocates rigs on a map, shows telemetry and QoE charts per channel and per device, and generates automatic alerts for conditions such as lost heartbeat, delayed telemetry, high temperature, weak WiFi signal or degraded playback quality.

RESULTS

The operation moved from complaint-based diagnosis to continuous evidence for each Dbox. Operations teams can see in real time the WiFi signal, system load, temperature, app stability and playback quality per channel. Degradations can be correlated with location, set-top box model, network conditions and time of day, enabling prioritized interventions based on concrete data and reducing detection time.

The platform runs on-premise inside the operator's network, with the option of managed SaaS deployment, and is ready to scale to thousands of devices distributed across any region.

← Back to success stories I want something similar